Semiautomated QT interval measurement in electrocardiograms from a thorough QT study: comparison of the grouped and ungrouped superimposed median beat methods

IntroductionWe postulated that it may be easier to identify earliest Q onset and latest T offset when the median beats from 12 leads are separated vertically by 5 to 10 mm (ungrouped superimposed median beat [SMB] method) rather than when their baselines closely (but rarely perfectly) overlap (grouped SMB method).MethodsThree readers manually adjudicated annotations placed by an automated algorithm, using grouped (gSMB) and ungrouped (uSMB) methods in 2658 electrocardiograms (ECGs) recorded in 38 subjects in a crossover design thorough QT study at predose and 6 time points postdosing with placebo or moxifloxacin.ResultsPlacebo-subtracted, moxifloxacin-induced QTcF prolongation was comparable with both methods. Maximum QTcF prolongation was seen at 2 hours—10.5 milliseconds (90% confidence interval, 7.9-13.1 milliseconds) with gSMB and 12.9 milliseconds (90% confidence interval, 9.9-15.8 milliseconds) by uSMB. Both methods showed good agreement; mean QT was 4 milliseconds greater by uSMB. Interreader variability of absolute differences in QT measurements was 1 millisecond lower with the uSMB method (6.8 ± 5.7 milliseconds by gSMB and 5.9 ± 4.5 milliseconds by uSMB).ConclusionMean QT was 4 milliseconds longer, and interreader variability, 1 millisecond lower with uSMB. Otherwise, both methods were comparable and detected the moxifloxacin effect.